Preliminary Design Review of the DUNE FD2-VD Photon Detector System (PDS)
Zoom: https://fnal.zoom.us/j/
Review Information
Review Home
Charge Letter
Tues May 3
11:30 AM
Stephen: Is Xe assumed in the calculation of the light yield?
Answer: yes
Bob: If you didn’t have Xenon what would the range be per MeV?
Answer: The main drawback without Xe is the light yield uniformity that you lose. Uniformity with Xe doping x2 to x3 maximum to minimum. Without Xe nonuniformity varies by an order of magnitude?
Jose: On the issue of non-uniformity - was the cathode mesh considered in the simulation? Is there dependence on the incident angle?
Answer: It is taken into account in the simulation - takes into account the presence of the mesh on top of the detectors located on the cathode. We define it around 90% geometric transperancy. Light yield takes into account also shadowing. Both reflectivity and transmission is taken into account in the MC.
Giovanna: Slide 7- you have PDS system that can see 10 interactions at low energy - do you have simulation also on the backgrounds at low energy?
Answer: It is in progress but we don’t yet have the results available. Main topic on how to separate backgrounds at low energy in trigger is active area of study.
Srini: Do the interface documents need to be agreed and signed off on by the relevant consortia?
Answer: Yes,
Srini: Do you have a risk register document?
Answer: There is a joint risk schedule held by the entire experiment
Srini: Good to have an estimated date of completion for the documents on slide 20
Jose: I didn’t see any documentation in the response monitoring system
Answer: Largely informed by FD1 Light Calibration Module - click link on slide 6
Jose, Stephen: Please add the documents linked on slide 6 to the summary sheet so we can include them in our review
Answer: Will do
Stephen: I don’t see any installation facility requirements - like space or lighting or for assembly
Answer: Some of that is addressed in HVS including processes outside the facility - we do have some of that in the installation documentation - that may be something that needs to be matured.
Bob: Are we going to see the fibres tested at full length in cold in high field conditions? Internet is full of people worrying about this.
Answer: Not full length or full field
Jose: On the issue of the LY requirements - what is the rationale to attribute some physics to the average LY requirements and other physics to the minimum LY requirements? For example minimum LY requirements are based solely on identification of backgrounds for proton decay but we do consider the background for identifying SN efficiently or not having the required energy resolution in the minimum LY requirements?
Answer: For identifying the background for SN - the TPC signal does well for that unless there is an issue with HV or purity since LY is less sensitive to the conditions than charge. For nucleon decay there is no TPC substitute for the t0 measurement required from the light system
Bob: Slide 7 what is the physics red curve?
Answer: This is the limit of the resolution given by the physics like visible energy
Bob: Is there a joint document with the LE SN group that they agreed that this is right requirement to have for the SN?
Answer: The requirements are developed jointly with the physics groups - particularly Alex Himmel.
Bob: Not recognizing the neutrons is a big problem with the HD not associate the right neutron vertex. The red curve physics limitation could actually be better since you have better chance here of detecting the neutron with the PDS. I am concerned that the requirement should be based on also looking at the detached neutron vertices as well. I am concerned finalizing the requirements before understanding better the red curve limitations.
Answer: The aim of the current simulation campaign underway is precisely to understand the actual physics resolution limitation.
Stephen: I keep hearing 4pi - are there reflectors on the cold electronics?
Answer: I believe there is some reflectivity there - around 50% for the wavelengths of interest. There are some recent studies with the CRP last layer in Lar with improved reflectivity on this last layer with different metals.
Stephen: What about the shielding/shadowing of the Sipm support?
Answer: That is a key tradeoff that we are going to be looking at
Gary: Slide 6 - I am trying to understand if there is specific choke points even though the sheet current is low there could be localized heating. I don’t see a thermal analysis for this current flow. How is the electro-mechanical interface to the modules are done?
Answer: Where the Arapuca is there is a conductive mesh above - we are not expecting local heating. We are only coupled to the mesh through high resistive loads - like 1 Mohm
Bob: When you do your testing - have you identifying a place where the discharge is more damaging than others?
Answer: We are looking at worst case in slide 6 - we treat the whole module as if it were receiving the uniform full current. The study shows that why we are not comfortable connecting conductively the first module.
Manhong: Whose scope is the mesh? For the membrane mounted module is there a mesh on both sides?
Answer: The mesh is HVS scope - for the membrane mounted we are behind the field cage there is no additional mesh around the Arapuca.
Stephen: Going from cold to warm - how is that accomplished? How does the signal get from inside to outside - what is the structure of the flange? Is the fibre continuous?
Answer: Slide 8: Viton cork with holes in it that then gets squished by metal parts and the fibre is continuous. It is the same thing used by the flashers in the purity monitor
Stephen: This has me worried - the optical fibre you are using has a cladding - the things used in PAB are not nominally used for optical fibre.
Answer: The ones we are using are actually used for optical fibre
Giovanna: Do you have the number of the various components for the VD? How many Daphne boards for example do you need?
Answer: One Daphne board per penetration and 40 penetrations = 40 boards
Carl: Some of opamps are bipolar you made the claim that hot carrier effects are not of concern - there is some literature coming out that indicate hot carrier effects are problematic even for bipolar opamps
Answer: The aging test for the bipolar amplifier has been carried out for the HD and we got a lifetime projection of over 20 years.
Bob: Are those in the documentation? The test results?
Jose: Do you have any requirements on uniformity of the various signal drivers or the whole chain from the laser driver.
Answer: We don’t have a requirement on that - we plan to calibrate all channels. If you can identify a single SPE peak. We are using similar pulser system to the horizontal drift to calibrate. LED flasher and diffuser was utilized in the cold box test to test linearity - see Sabrina's talk tomorrow.
Giovanna: In order to go very high in sensitivity you need to put the thresholds in the PDS low - is there an estimate of the data rates needed for this system through Daphne to the DAQ. This will have an impact on the requirement on the number of the Daphne boards
Answer: I don’t think we have made an estimate of the rates although I suspect it is low.
Gary: Coming back to the specification for the production spread (last line in the DUNE specs slide) - why is it 2V?
Answer: It is a compromise between can be realistically done and desirable
Bob: You thermal cycled the FPK sipms 20 times - is that for the whole batch?
Answer: Only prepoduction is cycled 20 times and production the SiPMs will be cycled 3 times both for FPK and Hamamatsu
Gary: Of the testing you have done so far - what is the failure rate you have seen?
Answer: Cryo reliability we have seen 2 failures of 2000 SiPMS from FPK and 2 failures from 1000 Hamamatsu SiPMS. The failure rate was primary in the increase of the dark count rate. We suspect this is because the thermal stresses are causing more surface leakage
Jose: What about long term tests?
Answer: We did this in Milano - we kept 24 SiPMs in LN temperatures for 2 months with no failures. Both types of SiPMS
Jose: With the gang schemes that you have tested - would you have any way of knowing that a particular SiPM has failed in the cold?
Answer: No - we only know that the channel has failed or we see the aggregate dark count rate has gone down.
Jose: In which case do you switch off the channel?
Answer: If we have a short we switch off a channel - 80 SiPMs. The channels are independent. We have separate bias in the channels. We should emulate the failure . The two channels are not optically seperated, but the readout is split into two separate readout channels.
Carl: You ran the SiPMS for 2 months. Any accelerated lifetime tests and any experience on lifetime in cold temperatures?
Answer: To the best of our knowledge the failure rate is dominated by thermal stresses so that is why we do the thermal cycling tests. We have not operated at overcurrent since the aging test is focused on thermal stresses.
Bob: You tested in the cold box at 10kV - how far from operations? I worry about the change in the reflectivity with different electric field.
Answer: We expected to go to 300kV - the tests showed there is complete electrical isolation and the absence of additional noise introduced by switching on and off HV system. Module 0 will be the test with the full HV and validation. We note the e-field is the same in both tests at 10kV test stand.
Manhong: You mentioned the WLS plate is surrounded by SiPM boards. When I looked at the 3-D model it only showed only side has a SiPM boards. The model doesn’t seem to have all the details yet
Answer: It is installed on all 4 sides 40 SiPMs per side for a total of 160. Each flex board has 20 SiPMS. We have 2 flex boards per side
Stephen: Can you indicate which 4 strips are ganged together to make a channel?
Answer: In the present prototype for simplicity there are 4 flex boards forming 1 channel what we are considering to alternate the SiPMS ganged together but the one that has been produced we don’t have this scheme implemented.
Gary: The step-up converter is very important - can we get more detail? I worry about noise if you are using magnetic components for the step-up convertors.
Answer: I started doing this in discrete components and built my own step-up components but we then found an off-the-shelf DC-DC convertor that worked. There is a discrete component version and off-the-shelf component that are being tested we don’t know which we will select. We add filters to address the noise issues.
Jose: How close are you to the limit of bubbling with the SiPM option? What is the main driving efficiency to go to the GaAr?
Answer: We saw no bubbling with the 40cm Lar test. We don’t really at high power with the SiPMs so we can run the lasers at higher power. You want redundancy as well as high efficiency so that is the reason for GaAr.
Stephen: Do you have an org chart that shows who is doing the different activities in the PoF group? What are the different roles do different players play - there is transverse effort on design and R&D and also different relationships vertically to deliver the product.
Answer: We have an org chart overall that shows the overall PDS effort we will present at the cost and risk.
Wed May 4
Jose: Time-over-threshold do you have enough bandwidth to measure that time with enough precision?
Answer: It is just an idea to solve the current linearity issue on slide 10. Concerning the precision- I am not sure what precision would be needed.
Stephen: Is there going to be any report on the wall mounted devices and what you learnt from them? Like the noise level comparisons?
Answer: Some studies on the noise levels with the detector on the wall powered with copper instead of POF and comparison with cathode are still ongoing. The noise is higher with the wall mount since it is at an earlier stage of prototyping and not as clearly defined.
Stephen: What is setting the bandwidth requirement (35 MHz) which seems high and makes
Answer: To transmit from SiPM signal we estimated we need 20 MHz and the limitation on the hardware is the first stage of the circuit.
Jose: On the issue of the SPE signals and you have to go looking by eye - how do you address this? Will you change something on the configuration?
Answer: The performance is good in the lab and there have been more fixes done in the cold box that will help. Also there has been some changes to the wall mounting shielding connections to help with the noise.
Alberto: What do you mean alternative designs are still in play? I know that there is an alternative design with the X-ARAPUCA on the wall.
Answer: No the X-ARAPUCA alternative on the walls is no longer in play. I am talking about alternatives in terms of components like fibres etc
Stephen: Slide 2 how would you answer the 40% completion question?
Answer: We have a preliminary design but I don’t think the design completion level is appropriate.
Gary: Compared to FD1 - what are the major differences/challenges? What about the membrane mount modules?
Answer: The wider plates need better mounting schemes for the SiPMs and wavelength shifting plates. The membrane
Mount modules actually draw on a lot of the experience with FD1 so I am not worried about that.
Albert: Where is the electronics mounted? How are they protected from HV?
Answer: Refer back to
Srini: Why are you going after multiple vendors for the dichroic filters when the ones at CD1 already worked?
Answer: There was interest in seeing whether we could improve performance - not clear yet. Also looking at glass substrates and making the larger plates. Try not to have a single vendor solution for any component that is the strategy to mitigate risk.
Manhong: The PD module is spring loaded. The membrane mount module is different - do you have a load case that considered spring loading in the simulation analysis plan?
Answer: No we have not included spring loading in the simulation. The WLS plates are actually guided on pins and keep the structure centered. You are right that when we go into vertical mounting the spring on the bottom is doing less and the top is doing more but the difference is small since the buoncy is helping. The WLS plates the load is not fully supported by the springs.
Stephen: Flavio told us we got a lot of PEs already - is that extracted from the simulation or is it from a measurement? If it is from measurement why do you want more given finite amount of time and resources?
Answer: The efficiency in the photocollection of the large surface ARAPUCA has not been measured yet simply because that is not possible in the timeframe so far. There are however a number of increasingly detailed simulation that essentially confirm the expectation of the efficiency that we have for these modules - in the range 2-3% of overall efficiency (detected photons/total photons impinging on surface). The measurements with ARAPUCA from previous designs agreed with the simulation. Verification of the large area surface devices collection efficiency is planned in the last cold box of this year or early next year - but not the next cold box test.
Jose: what is the efficiency for light collection at higher angles than 45 degrees - say grazing?
Answer: We lose efficiency - which is a motivation of adding more dichroic filters in the Italian devices. But it may not be necessary since the indication from the simulation where we did look at the simulation efficiency at different incident angles is that we have sufficient light.
Thursday May 5
Stephen: What is construction on slide 3?
Answer: Construction on bottom row of the table is for Module 0 then production refers to full DUNE production
Srini: Slide 4: why 9% - strange number? Does this cover the yield? Odd to have 9% for all systems (slide 5)? Don’t different systems have different yields like SiPMs where you have 20% spares.
Answer: The 9% covers the expected yield
Alberto: You may need new complete modules if you have failures in the installation stage.
Alberto: Do you test all the modules?
Answer: Modules are assembled and tested then on a table in the clean room on the surface then installed on the cathode. We need to solve the problem with testing warm with a class 4 laser with personnel around. Testing the module is quick but the challenge is having to use the class 4 lasers
Stephen: Slide 3 - when is the FDR?
Answer: Goal is January 2023
Srini: It is important to note that the input to the FDR is the cold box tests and the module 0 tests is before the PRR and after the FDR
Answer: The module 0 plan shown here has an acceleration of 6 months from a couple of months ago.
Stephen: Why do you mean by EU? Unless the funding is from the European Union specifically it is a misleading abbreviation. It is better to use Europe or Eu.
Answer: Spain/Italy (INFN)/France
Alberto: I added up the costs of the excel spreadsheet on EDMS for both cathode and membrane and they similar to 10% - how can this be given that the cathode needs POF and dichroic filters on both sides for example?
Answer: We need to produce an accurate cost estimate separately for cathode and membrane components from p6.
Srini: In the highest level risks what is your strategy to mitigate the risks?
Answer: For example the simulation shows additional detectors are needed then the response is to add more modules.
Srini: So what is the mitigation of that risk as opposed to the response?
Alberto: That is why it is important to understand the cost of ARAPUCA's on the membrane vs cathode to understand how to mitigate that risk since if you have pro
Stephen: How come that doesn’t have a risk on the schedule?
Answer: That is because we have enough float in the schedule. We would spend money on shifts and labor to avoid schedule impact.
Carl: Have you thought about the risk of not being able to procure the electronics components on time? Like we have run into issues with ordering opamps necessitated redesign of components.
Answer: We tried to mitigate risks using two vendors - our components are standard commercial and we have several opamps tested at BNL that could be substituted - this should be about 20-30% probability risk. We don’t have risk for electrical components like we do for mechanicals.
Jose: What about risks for component damage during installation? Fibres are fragile for example. Are these considerations taken into account?
Answer: There are project wide risks for shipping and handling.
Srini: What about the risk of partners not delivering or not delivering on time - what is the risk to US scope?
Answer: We proposed this risk and it was rejected.
Srini: What about risks on interfacing systems where risks on them have impact on you or vice versa?
Answer: We do have risks for interfacing with installation and the detector generates noise on the TPC strips
Alberto: For the shipping I don’t think there is a high level risk in the project.
Answer: The dichroic filters are very delicate, but I recall at the last project meeting they added a project while risk
Stephen: It will be worthwhile to retire the risks you can retire - or make it clear in the slide which risks are retired (strikethrough)
Answer: The top risks are retired
Stephen: The project should take on insurance risk on a project wide basis
Alberto: We were told Fermilab would not pay for insurance.
Stephen: This is a project wide issue.
Stephen: I don’t understand the bias in who comes up with funding - this is a general risk not just a risk to DOE alone.
Answer: We have a p6 schedule for nominally an international project but it is really a US project function so we don’t include risk to the French project for example.
Srini: Is there an excel spreadsheet to maintain risks on the international side?
Answer: Our L2 rebels when asked to keep two separate registries for US and International partner risk so each national project keeps its own risk. We cant comment on the quality of these efforts.
Stephen: What item here addresses funding risks from the DOE?
Answer: We are struggling with a project that has baselined parts of the project and we are at pre-CD1 level.
Stephen: It raised my ire (??) that we only consider international schedule and deliverable risk but not the DOE funding and schedule delays
Srini: In answer to Stephen the reason we are focused on the international risk to schedule and funding is because we have a process established on the US project side but not a similar process on the international side
Stephen: I have seen such risks from DOE funding from Eric James in other parts of the project
Bob/Srini: I suspect that is because these risks are held at the overall FD2 level not at the PDS level and as such are not appropriate at this level.
Bob: What are the requirements to estimate scope contingency items?
Alberto: What type of contingency do you have for the PDS?
Answer: There is a line by line estimate uncertainty on the subsystem level that is fed into the contingency
Mary: What recommendations would you like from the committee?
Answer: Very valuable if there was recommendation that the cold box schedule the PD priority it cant be CRP setting all the scheduled. The mechanical concepts need to better co-ordinated and the models need to be updated in a coherent way on a short time-scales. Our fibre installation affects every other subsystem. All the models are updated at different times and different timescales. There is a competition/co-ordination within the PDS consortia needed between FD1 and FD2. Some of the EU groups are scrambling to get money needed to address the schedule acceleration by 6 months mentioned earlier.
Steve: How has the PDS consortia reacted to the acceleration in schedule and struggle for funding in terms of trying to recruit new effort to address the extra resource need for 2022?
Alberto: For the interfaces for FD1 we had one person taking care of the interfaces and a process for resolving issues. For FD2 we don’t have such a person or a repository system for co-ordinating this.